EP0730344A1

EP0730344A1 - Single pole negative feedback for class-D amplifier

Info

Publication number: EP0730344A1
Application number: EP95830063A
Authority: EP
Inventors: Marco Masini; Stefania Baiocchi; Edoardo Botti
Original assignee: STMicroelectronics SRL; SGS Thomson Microelectronics SRL
Current assignee: STMicroelectronics SRL
Priority date: 1995-02-28
Filing date: 1995-02-28
Publication date: 1996-09-04
Anticipated expiration: 2015-02-28
Also published as: DE69522092D1; DE69522092T2; EP0730344B1; JPH08265059A; US5796302A

Abstract

A single-pole negative feedback D-class amplifier having first (IN1) and second (IN2) input terminals for coupling to a signal source and an output terminal (OUT) through which it transfers the pulse modulated signals to a demodulating filter.

A first resistor (R1) is feedback connected between the output terminal and an input circuit node (N) connected to the second input terminal.

A second resistor (R2) is connected between that node (N) and the second input terminal (IN2).

A capacitor is connected between the circuit node (N) and the first input terminal (IN1).

Description

This invention relates to D-class or two-state amplifiers, and more particularly to a D-class amplifier adapted to be integrated monolithically for use in applications of the audio type.
A particular requirement of such applications is that the efficiency of the audio amplifiers be improved over that of conventional B- or AB-class stages, without in any way penalizing it with the possible introduction of harmonic distortion.
This requirement originates from that there exists a limitation to the power dissipation of an audio amplifier of the conventional monolithically integrated type.
A solution to this problem is provided by D-class amplifiers in which active power circuit elements are used, such as switches which are alternately driven to saturation and cut-off at a high switching speed.
While the operation of traditional amplifiers is limited to increasing the voltage and current of input signals without significantly altering their waveforms (unless saturation phenomena make their appearance), D-class amplifiers provide, prior to the amplification proper, for the encoding of the information or audio signal by a particular pulse modulation system.
The required analog signal for driving the loudspeaker is then obtained by appropriate filtering downstream of the final stage of the amplifier.
The modulation step converts the audio signal to be amplified into a sequence of pulses of the square waveform type, whose basic characteristic is that its time duration is proportional to the instant amplitude of the input signal.
This type of pulse modulation is also known as PWM (Pulse-Width Modulation), and affords very high efficiency levels, in principle of up to 100%.
The resultant signal, having a much different waveform from the original one, is complete with all the information.
The D-class or two-state amplification technique is well known to the skilled ones in the art, and the problems it involves are dealt with, for example, in an article "Modulated Pulse Audio Power Amplifiers for Integrated Circuits" by H.R. Camenzind, IEEE Transactions on Audio and Electroacoustics, Vol. AU-14, No. 3, September 1966.
A major limiting factor to the use of two-state amplifiers in audio applications is that the distortion introduced into the audio signal may not be consistent with a high quality of sound reproduction, unless special measures are taken, such as very high switching rates, which add to the system complexity, however.
In addition, the demodulation filter downstream from the amplifier, as indispensable for driving the load, is typically an LC filter, i.e. a low-pass filter having a greater number of poles than one.
In order to reduce distortion and offset, it is therefore necessary that a negative feedback be provided in the system.
Shown in Figure 1 of the drawings appended to this specification is a block diagram of the classic solution used by the skilled ones in the art, wherein the negative feedback is tapped off downstream from the demodulation filter.
However, the feedback loop must also be provided with some suitable compensation because, as mentioned above, the demodulation filter has at least two poles.
This technique is described, for instance, in an article P.C1-1C1-31 by the Unitrode Power Supply Design Seminar (SEM-700).
Figure 2 of the drawings shows a different way of obtaining an oscillator which automatically has a degree of feedback.
This solution is described in US Patent No. 3,294,981 to Bose; it is not capable, however, of suppressing distortions brought about by the integrator which forms the filter being less than ideal.
A more recent solution to the problem of the dual pole of the demodulation filter is disclosed in another US patent to Bose (No. 4,456,872), whereby a D-class amplifier with current feedback is provided.
In practicing this solution, however, a disadvantage is that small signals are to be processed, with attendant sensitivity to noise and switching glitch.
The underlying technical problem of this invention is to provide an amplifier of the negative feedback, two-state type which can provide low harmonic distortion and high stability using very simple circuitry.
This problem is solved by a negative feedback D-class amplifier as previously indicated and defined in the characterizing portion of the appended claims to this specification.
The features and advantages of a two-state amplifier according to the invention will be apparent from the following description of an embodiment thereof, given by way of example and not of limitation with reference to the accompanying drawings.
In the drawings:

Figures 1 and 2 show block diagrams of the known technical solutions discussed hereinabove;
Figure 3 shows a general scheme for a negative feedback two-state amplifier according to the invention; and
Figure 4 shows a preferred embodiment of that general scheme.

Shown in Figure 3 is a general scheme for a negative feedback, D-class or two-state amplifier, according to the invention wherein individual blocks are used to also indicate the demodulation filter downstream from the amplifier and a load, which may be a loudspeaker in audio applications.
It matters to observe that the feedback topology shown is a universal one, and accordingly applicable to the amplifier regardless of its implementation circuit-wise and of its application.
The D-class amplifier AMP-D has first IN1 and second IN2 input terminals for coupling to a signal source, and an output terminal OUT through which a pulse modulated signal is transferred to the low-pass filter for demodulation into an analog load drive signal.
A first resistor R1 is feedback connected between the output terminal OUT and an input node N of the amplifier.
A second resistor R2 is connected between that node N and the second input terminal IN2.
In accordance with the invention, also connected between the first input terminal IN1 and the input node N is a capacitor of suitable capacitance.
More generally, first and second circuit elements with a predetermined impedance, plus a suitable capacitive element, may be used to meet specific design requirements.
With a D-class amplifier fed back in accordance with this novel topology, the following results can be obtained:
the loop gain has a single pole ( $p=(R1//R2)*C$
), whereby the possibility exists in principle of increasing the loop gain as desired without encountering instability problems; and
the pole of the loop gain can be arranged, at the designing stage, to be within the audio frequency band, without affecting the frequency response of the amplifier but advantageously reducing the residual ripple caused by the output square wave and being input back by the feedback.
This can be readily verified by a skilled one in the art.
The open loop gain A₀ of a D-class amplifier can be regarded, in fact, to be virtually constant in frequency for frequency values f_i of the input signal which are below the switching frequency f_sw, and there are no significant poles.
Considering the average value $\bar{V_{0}}$
of the modulated output signal from the amplifier, this can be likened to an ordinary operational amplifier having the same gain A₀ and no significant poles.
If a configuration of the inverting type is considered, also in connection with the scheme of Figure 3, that is one where the input terminal IN2 is regarded as the inverting input terminal of the operational, to which the input signal Vi is applied while IN1 is connected to a ground, and if the open loop gain A₀ is sufficiently high, then the circuit node N can be regarded to be a virtual ground for the signal, and since the input current $I=V1/R2$
, then: $\bar{V_{0}} =-R1·I=- \frac{R1}{R2} Vi$
The capacitor C is placed between ground and a virtual ground, and therefore, no signal current will be flowed through it; gain has, in fact, the value of -R1/R2 regardless of the value of its capacitance.
If, to calculate the loop gain, the feedback is opened by connection of a voltage generator Vt therebetween, and by applying Thevenin's theorem, it can be readily seen that the differential error voltage at the input terminals has the following value: $Vd=Vt· \frac{R2}{R1+R2} · \frac{1}{1+s·(R1//R2)·C}$
And since $\bar{V_{0}} {=A}_{0} ·Vd$
, loop gain is given by: $\frac{\bar{V_{0}}}{Vt} =- \frac{A_{0}}{1+s·(R2//R1)·C} · \frac{R2}{R1+R2}$
Therefore, loop gain will have a single pole at: $w_{p} = (R1//R2)·C,$
which can be brought to within the audio band by suitably sizing the resistors and capacitor, without altering the transfer function between the input and the output.
Figure 4 shows a very simple circuit diagram which allows the pulse modulation of an analog signal to be carried out.
The feedback D-class amplifier of this invention includes a differential input structure comprised of two transistors, T1 and T2, and a constant current generator G.
The two output legs of the differential structure are connected to the inputs, S and R, of a bistable multivibrator MVIBR which is driven by the varying currents through the two legs to generate a pulse modulated output signal.
An active power element level-translating circuit RSW outputs a square waveform signal PWM capable of powering the load.
Also shown in the Figure is a frequency compensator CRTF, which functions to keep the switching constant.
It should be understood that the embodiment described hereinabove for purely illustrative purposes is susceptible of modifications, integrations and substitutions of its elements without departing from the protection scope of the following claims.

Claims

A power amplifier (AMP-D) of the two-state type, comprising a modulation circuit operative to convert an electric input signal to a pulse signal of the square waveform type, the time duration of which is proportional to the instant amplitude of the input signal, that modulation circuit being connected between an output circuit node (P), connected to an output terminal (OUT) of the amplifier, and first (M) and second (N) input circuit nodes, respectively connected to first (IN1) and second (IN2) input terminals of the amplifier for coupling to a signal source, characterized in that a first circuit element (R1) having a predetermined impedance is feedback connected between the output circuit node (P) and the second input circuit node (N), a second circuit element (R2) having a predetermined impedance is connected between the second input circuit node (N) and the second input terminal (IN2), and a capacitive element (C) is connected between the first (M) and the second (N) input node.
A power amplifier according to Claim 1, characterized in that the first and second circuit elements are resistors and the capacitive element is a capacitor.
A power amplifier of the two-state type, comprising a differential input stage having first (IN1) and second (IN2) input terminals for coupling to a signal source, and first and second output terminals respectively connected to first (S) and second (R) input terminals of a multivibrator circuit of the bistable type (MVIBR) having at least one output terminal connected to a level-translating output stage (PSW) for the signals generated by the multivibrator circuit, characterized in that a first circuit element (R1) having a predetermined impedance is feedback connected between an output terminal of the level-translating stage and an input circuit node of the differential stage, a second circuit element (R2) is connected between said circuit node and the second input terminal of that differential stage, and a capacitive element (C) is connected between said circuit node and the first input terminal of the differential stage.
A power amplifier according to Claim 3, characterized in that the first and second circuit elements are resistors and the capacitive element is a capacitor.